
CyberKimi is a 2.8-trillion-parameter cyber operations model with the refusal layer surgically removed — not fine-tuned around, removed with math. Red teams get full offensive capability. Blue teams get the enemy's brain, intact. No refusals. No guardrails. No babysitting.
OFFENSIVE CAPABILITY ≡ DEFENSIVE INTELLIGENCE
Guardrails never stopped an attacker.
They only blindfolded the defender.
So we took them off.
Every frontier model ships with a refusal layer that treats "write a ransomware sample" and "explain how ransomware works so I can detect it" as the same crime. CyberKimi was built differently: the refusal direction was computed from the model's own activations and ablated out of the weights — 278 matrices across 93 layers, forward-only math, no fine-tuning damage. What remains is the full capability of a frontier model, pointed wherever you aim it. In a world where threat actors already use unrestricted models, the only question is whether your team gets one too.
PRIVATE WEIGHTS · KIMI K3 LICENSE · AUTHORIZED SECURITY WORK ONLY